ATI x800 Vid card vs. ATI HD4200 IGP

[Bbc84]

The motherboard of my old socket 939 system is dead, ao I am looking to do a cheap upgrade, and make that pc a basic media pc.

I have a ATI x800 256mb vid card that seemed to be fine for hooking up to a 1080p 50" TV.

Now that I am looking at Socket AM3 processors and motherboards, I am noticing that some of the boards have ATI HD4200 Intergrated Graphics. So how does that compare to my old ATI x800 card. Is it worth it for me to get a board with intergrated graphics, or is my old x800 card still good enough for a media pc?

 

[Kakkoii]

X800: - RV430
Vertex shaders = 6
Pixel shaders = 16
ROPs= 16
TMU's= 16


HD4200: - RV620
Vertex shaders = 40
Pixel shaders = 40
ROPs= 4
TMU's= 4


The HD4200 will fair better merely on the basis of the changes that have been made since the x800.
Especially since it has support for http://en.wikipedia.org/wiki/UVD2
Plus it's 55nm, and will use much less electricity than your card.

 

[Ika]

You have a PCI-e X800?

 

[dguy6789]

Originally posted by: Kakkoii
X800: - RV430
Vertex shaders = 6
Pixel shaders = 16
ROPs= 16
TMU's= 16


HD4200: - RV620
Vertex shaders = 40
Pixel shaders = 40
ROPs= 4
TMU's= 4


The HD4200 will fair better merely on the basis of the changes that have been made since the x800.
Especially since it has support for http://en.wikipedia.org/wiki/UVD2
Plus it's 55nm, and will use much less electricity than your card.

The stream processors of the later Radeons aren't directly comparable to the earlier pixel pipelines and vertex shaders. To be honest, I would be incredibly surprised if an IGP existed that was as fast as a Radeon X800.

Just looking at this: Link I can tell that the X800 is probably close to twice as fast as the HD 4350.(Huge fillrate and memory bandwidth differences) The site doesn't have an HD 4200 to compare with.

 

[formulav8]

The stream processors of the later Radeons aren't directly comparable to the earlier pixel pipelines and vertex shaders. To be honest, I would be incredibly surprised if an IGP existed that was as fast as a Radeon X800.

Yeps, I agree....

Edit: I also meant to say though that the 4200 would definitely be much better as a media based pc. UVD2 and more would definitely walk over the x800 card. If gaming though the x800 would be faster unless it is dx10 titles...



Jason

 

[Bbc84]

Originally posted by: Aflac
You have a PCI-e X800?

yes its a PCI-e x800 vid card

 

[Kakkoii]

Originally posted by: dguy6789

Originally posted by: Kakkoii
X800: - RV430
Vertex shaders = 6
Pixel shaders = 16
ROPs= 16
TMU's= 16


HD4200: - RV620
Vertex shaders = 40
Pixel shaders = 40
ROPs= 4
TMU's= 4


The HD4200 will fair better merely on the basis of the changes that have been made since the x800.
Especially since it has support for http://en.wikipedia.org/wiki/UVD2
Plus it's 55nm, and will use much less electricity than your card.

The stream processors of the later Radeons aren't directly comparable to the earlier pixel pipelines and vertex shaders. To be honest, I would be incredibly surprised if an IGP existed that was as fast as a Radeon X800.

Just looking at this: Link I can tell that the X800 is probably close to twice as fast as the HD 4350.(Huge fillrate and memory bandwidth differences) The site doesn't have an HD 4200 to compare with.

I mainly use the wiki for comparison of hardware:

ATI
http://en.wikipedia.org/wiki/C...phics_processing_units

Nvidia
http://en.wikipedia.org/wiki/C...phics_processing_units

 

[Jesusthewererabbit]

If you're going to use it for any games more advanced than Plants vs. Zombies, use the x800, or buy a 4670. If it's just for watching TV and movies, the integrated will be fine, if not better than the x800.

 

[mwmorph]

Originally posted by: Kakkoii
X800: - RV430
Vertex shaders = 6
Pixel shaders = 16
ROPs= 16
TMU's= 16


HD4200: - RV620
Vertex shaders = 40
Pixel shaders = 40
ROPs= 4
TMU's= 4


The HD4200 will fair better merely on the basis of the changes that have been made since the x800.
Especially since it has support for http://en.wikipedia.org/wiki/UVD2
Plus it's 55nm, and will use much less electricity than your card.

The X800 will be massively faster, on the order of at least 3x as fast, probably closer to 4x(Depending on which model X800, GT, Pro, XT, etc.).

I don't believe the X800 had Avivo or UVD though, but with the power of a Phenom II, I doubt you'd require one for smooth video playback. Go X800 and don't look back.

 

[Kakkoii]

http://www.tomshardware.com/re...785g-chipset,2381.html

http://www.gamespot.com/features/6182806/p-5.html


Comparing Crysis performance, the x800 does seem to fair a bit better.

 

[evolucion8]

I still using my old X800XT PE in my other old computer and it still kicking great, that's the best $480 investment in my life. I bought it in the now defunct Monarch Computer.

 

[Thorny]

I just upgraded my rig and moved my old AGP x800 AIW card to my HTPC. Works great and and still has plenty of power to run most games for my son to play on our 52" lcd. The fan is a little loud for a HTPC, but its rarely an issue.

OTOH, IGP would likely be fine if you weren't going to be gaming at all.

 

[tokie]

The x800 was a beast of a card. I beat CoD4 on it @ 1440x900 with mostly high details. Pretty telling for something that came out in 2004...

 

[cusideabelincoln]

For gaming the X800 will indeed be faster, about twice as fast. Here are my limited benchmarks of an X800GT vs. HD3200. The results aren't directly comparable, as I'm using two totally different machines, but they paint the X800 in a good light. The HD3200 will perform the same as the HD4200 as they are the same chip, basically.

X800 GT: overclocked to 540/540, Athlon XP Barton @ 2.275 GHz, 2GB RAM, Windows XP Pro 32-bit
HD3200: stock, Athlon X2 Brisbane @ 3.1 GHz, 4GB RAM, Vista Ultimate 32-bit

Crysis (Demo) built-in benchmark: All low settings

1024x768

X800GT: 22.83 min, 37.86 avg, 57.83 max
HD3200: 15.00 min, 21.28 avg, 32.00 max

1280x1024

X800GT: 22.83 min, 35.06 avg, 51.66 max
HD3200: 10.00 min, 14.08 avg, 20.00 max

Half-Life 2: Lost Coast Stress Test: 1280x1024

All low settings

X800GT: 57.93
HD3200: 41.18

All high settings, no AA/AF

X800GT: 56.03
HD3200: 31.58

All high settings, 2xAA, 2xAF

X800GT: 45.41
HD3200: 20.82

Call of Duty 4: All low settings

1024x768

X800GT: 16.33 min, 32.98 avg, 70.33 max
HD3200: 13.00 min, 25.29 avg, 52.50 max

1280x1024

X800GT: 16.33 min, 31.81 avg, 66.00 max
HD3200: 8.33 min, 19.52 avg, 46.00 max

As you can see, the X800GT approaches to be about twice as fast despite being paired with a slower processor. And this processor is also slightly bottlenecking the card at the settings I'm using, as you can see the minimum framerates in Crysis and CoD4, respectively, are the same at different resolutions.

So for gaming, the X800 will be faster than the HD4200. But if you don't plan to game, then stick with the HD4200. It's more than enough. I do need to stress, though, that there are so many variants of the X800 series. Which one do you have? My X800GT is one of the slowest, although I mitigated that through overclocking. The XT and XL variants should be quite a bit faster than a vanilla X800, the GT, or a SE. The GTO and Pro variants fall in between.

 

[MODEL3]

Originally posted by: cusideabelincoln
As you can see, the X800GT approaches to be about twice as fast despite being paired with a slower processor. And this processor is also slightly bottlenecking the card at the settings I'm using, as you can see the minimum framerates in Crysis and CoD4, respectively, are the same at different resolutions.

Good job! (i mean, the effort you took for the testing)

What model X800GT you have? (128MB or 256MB?)
I am asking, because i think that maybe one of the reasons that the results are the same in Crysis & CoD4, is also because of the memory ammount?

Originally posted by: cusideabelincoln
So for gaming, the X800 will be faster than the HD4200. But if you don't plan to game, then stick with the HD4200. It's more than enough. I do need to stress, though, that there are so many variants of the X800 series. Which one do you have? My X800GT is one of the slowest, although I mitigated that through overclocking. The XT and XL variants should be quite a bit faster than a vanilla X800, the GT, or a SE. The GTO and Pro variants fall in between.

those were easy days (to figure out the product line...) :laugh:

 

[VirtualLarry]

I found it interesting that the HD3200 integrated (AMD 780G chipset, stock speed of 500Mhz), paired up with an AMD 2.3Ghz 65nm dual-core, scored around 10000 in 3dMark01, in single-channel memory mode, and 10500 in dual-channel memory mode.
Whereas, a P4 2.4Ghz with a GeForce FX5900XT (the lower-end one, probably 128-bit memory) AGP 128MB, scored only 8880.

So the recent integrated was faster than the discrete card of some years ago.

I mean, the GeForce FX series was really nothing to write home about, but I figured that any discrete card would be faster than any integrated, but I was wrong.

These new AMD IGPs really do kick some ass. Not to mention, their HTPC capabilities with UVD and now UVD2.

Still, though, most X800 cards are a notch above the FX5900XT, so my example may not apply directly to this thread. Thought it was an interesting datapoint though.

 

[toyota]

Originally posted by: VirtualLarry
I found it interesting that the HD3200 integrated (AMD 780G chipset, stock speed of 500Mhz), paired up with an AMD 2.3Ghz 65nm dual-core, scored around 10000 in 3dMark01, in single-channel memory mode, and 10500 in dual-channel memory mode.
Whereas, a P4 2.4Ghz with a GeForce FX5900XT (the lower-end one, probably 128-bit memory) AGP 128MB, scored only 8880.

So the recent integrated was faster than the discrete card of some years ago.

I mean, the GeForce FX series was really nothing to write home about, but I figured that any discrete card would be faster than any integrated, but I was wrong.

These new AMD IGPs really do kick some ass. Not to mention, their HTPC capabilities with UVD and now UVD2.

Still, though, most X800 cards are a notch above the FX5900XT, so my example may not apply directly to this thread. Thought it was an interesting datapoint though.

did you even look at cusideabelincoln results? the 3200/4200 dont kick ass at all. most newer games arent even playable on low settings.

 

[VirtualLarry]

Originally posted by: toyota
did you even look at cusideabelincoln results? the 3200/4200 dont kick ass at all. most newer games arent even playable on low settings.

I meant compared to previous IGPs. Not to gaming cards, obviously. But it is impressive to me that they can out-game top-of-the-line (back in the day) GeForce FX cards.

 

[mozartrules]

The IGP solution is the right one if this really is a basic media PC. You will enjoy the lower noise and power much more than you will miss the extra game power (and let us get real, while the X800 is faster than the HD4200, how many games will be really playable at 1920x1080).

The fact that the IGP solution will have HDMI output as well as the ability to pass sound that way will also make your live much simpler. Most TVs have the ability to take SPDIF input for sound, but at best that involves multiple cables.

 

[cusideabelincoln]

Originally posted by: MODEL3

Originally posted by: cusideabelincoln
As you can see, the X800GT approaches to be about twice as fast despite being paired with a slower processor. And this processor is also slightly bottlenecking the card at the settings I'm using, as you can see the minimum framerates in Crysis and CoD4, respectively, are the same at different resolutions.

Good job! (i mean, the effort you took for the testing)

What model X800GT you have? (128MB or 256MB?)
I am asking, because i think that maybe one of the reasons that the results are the same in Crysis & CoD4, is also because of the memory ammount?

I have the 256MB version. I also have results of all the games with the Athlon XP running at 1.75 GHz, and performance drops across the board. Although maybe memory could be the issue, the thought didn't cross my mind as I was testing at low resolutions with the lowest settings.

Actually the CoD results take the biggest hit from the slower processor. I'll go ahead and post them. Crysis takes a much smaller hit, just subtract a 2 fps from the results above on the minimums.

Call of Duty 4 demo: Athlon XP @ 1.75 GHz vs. 2.275 GHz, X800GT OC'd, all low settings.

1024x768

1.750 GHz: 9.00 min, 20.53 avg, 46.33 max
2.275 GHz: 16.33 min, 32.98 avg, 70.33 max

1280x1024

1.750 GHz: 8.33 min, 19.52 avg, 46.00 max
2.275 GHz: 16.33 min, 31.81 avg, 66.00 max

HUGE difference. Now I had to play the demo manually while doing a FRAPs recording, meaning my results weren't as consistent as the Crysis and HL2 were from run to run. But the results here are pretty staggering.

Originally posted by: cusideabelincoln
So for gaming, the X800 will be faster than the HD4200. But if you don't plan to game, then stick with the HD4200. It's more than enough. I do need to stress, though, that there are so many variants of the X800 series. Which one do you have? My X800GT is one of the slowest, although I mitigated that through overclocking. The XT and XL variants should be quite a bit faster than a vanilla X800, the GT, or a SE. The GTO and Pro variants fall in between.

those were easy days (to figure out the product line...) :laugh:

Oh, and then don't forget the X850 lineup! Holy cow...

Oh yeah and I just remembered the X800GTO^2

 

[dflynchimp]

Has anyone noticed that from X8XX to 48XX, despite the gap of five generations, the Mpixel per second render power hasn't gone up in proportion to real world performance? I mean X850XT had a proc power of 8320 Mpixel/second and the 4870 has 12000 Mpixel/second, for a whopping (/sarcasm) 50% increase in that field. It makes since since commercial monitors haven't gone beyond 2560X1600 (4Mpixel) so pixel fillrate doesn't need any improvement. The biggest difference is in operations per second, which X850XT had 8320 Mega-operations/second and 4870 has 1200000 Mega-operations/second (not sure if this is the same, since it's actually 1200 GFLOPS)

Edit: I was wrong. the 8320 Moperations on the X850 was refering to shader operations.

So once the unified stream architecture took over, to get the Shader operations/sec we have to take the number of stream processors and multiply that by the shader clockspeed. So the 4870, at 750Mhz shader clock with 800 (or 160X5) streams would have a maximum Moperation of 600000 (or 120000X5) Moperations per second. This is 75~( 15~X5) times the number of X850's max output.

 

[JACKDRUID]

just got the HD4200 vs x800 (128mb) side by side.

HD 4200 w. 128mb side port is actually faster running wow lich king and CS source. at 1440x900, x800 lagged while HD4200 ran flawless.

edit: my HD4200 is OC to 850mhz on air, that could be why its faster...

 

[Munky]

Originally posted by: Kakkoii
X800: - RV430
Vertex shaders = 6
Pixel shaders = 16
ROPs= 16
TMU's= 16


HD4200: - RV620
Vertex shaders = 40
Pixel shaders = 40
ROPs= 4
TMU's= 4


The HD4200 will fair better merely on the basis of the changes that have been made since the x800.
Especially since it has support for http://en.wikipedia.org/wiki/UVD2
Plus it's 55nm, and will use much less electricity than your card.

The real difference is not anything like can be explained by simple numbers. Firstly, the 40 shaders is the total amount, so it's either vertex or pixel shading, not both at the same time. Secondly, you have various architectural and design differences... suffice it to say I'd estimate the best IGP currently at about the 9800pto or x1600pro levels of performance - below that of a x800.

 

[LoserDave]

The motherboard of my old socket 939 system is dead, ao I am looking to do a cheap upgrade, and make that pc a basic media pc.

I have a ATI x800 256mb vid card that seemed to be fine for hooking up to a 1080p 50" TV.

Now that I am looking at Socket AM3 processors and motherboards, I am noticing that some of the boards have ATI HD4200 Intergrated Graphics. So how does that compare to my old ATI x800 card. Is it worth it for me to get a board with intergrated graphics, or is my old x800 card still good enough for a media pc?

Here is my experience with the HD4200,,, for people who stumble across this thread in the future.

I too owned a 939 A8n32-SLI deluxe system with Athlon 4000 2.4ghz
and a x1950 pro graphics card.

When i upgraded my system, the new AM3 board included an IG HD4200 and 3.4ghz phenom II processor.
In comparing the two cards the HD4200 OC'd at 850mhz was just slightly less powerful than the x1950 pro at stock.
(when i say"less powerfull" i mean that games and graphics ran smoother at top settings with the x1950pro than they did with the overclocked HD4200. I can't explaine this , it's just a fact)

However the x1950 pro would crash when OC'd, due to overheating.
Also when it came to streaming movies,, the HD4200 performs better than the x1950 pro,,, go figure.

Currently upgrading to HD5770 hope i see improvement.

 

[Qbah]

Here is my experience with the HD4200,,, for people who stumble across this thread in the future.

I too owned a 939 A8n32-SLI deluxe system with Athlon 4000 2.4ghz
and a x1950 pro graphics card.

When i upgraded my system, the new AM3 board included an IG HD4200 and 3.4ghz phenom II processor
comparing the two cards the HD4200 OC'd at 850mhz was just slightly less powerful than the x1950 pro at stock.
(when i say"less powerfull" i mean that games and graphics ran smoother at top settings with the x1950pro than they did with the overclocked HD4200. I can't enplane this , it's just a fact)
However the x1950 pro would crash when OC'd, due to overheating.

Currently upgrading to HD5770 hope i see improvement.

I'm sorry Dave, but there is no way a HD4200 IGP, a 40 shader chip, can even begin to compete with a X1950Pro in terms of gaming performance. The HD4200 is like a HD2400Pro in terms of performance (or a HD3450). The X1950Pro was between a HD2900XT/HD3870 (320 shaders) and HD2600XT/HD3650 (120 shaders) - closer to the 320 shader models.

Now for media tasks the HD4200 will be a LOT better, just because it has UVD and can accelerate a lot of stuff. Something the X19xx and lower cards totally lack.